SL Baur <steve(a)xemacs.org&gt; writes: > Here's a sledgehammer test to detect stuff like this in the future (no > doubt there's room for optimzation, this is slow and pure brute force): > > (mapatoms > (lambda (symbol) > (cond > ((boundp symbol) > (prin1 (symbol-name symbol)) (terpri) > (prin1 (symbol-value symbol)) (terpri)) > ((fboundp symbol) (symbol-function symbol)) > (t nil)))) > > 1998-07-27 SL Baur <steve(a)altair.xemacs.org&gt; > > * keymap.c (vars_of_keymap): Initialize Vkey_translation_map and > Vvertical_divider_map. Isn't this something that could just be done in DEFVAR_LISP? Slap a *c_location = Qnil; at the end of DEFVAR_LISP in symeval.h? Then you would only ever have to worry about explicit initialization if it needed to be something other than Qnil. Similar for DEFVAR_INT, DEFVAR_BOOL, etc. -Bill P.